Previous blog shows how to Add Attachments and Pictures to Azure DevOps Tasks from Email
There might be a need for adding some AI to the solution – why not use AI Builder to help out with the root cause analysis and solution proposal. This is very easily added to the solution once there is everything else in place.
Go to AI Hub, you can find it behind the More button in navigation and then Discover all button
Custom Prompts
Create a custom prompt for ChatGPT prompt with parameters for seeking solution. These are just for demonstrating purposes, create better ones that suits for your need.
And another for analysing the error
Publish the prompts and change to Power Automate flow
Power Automate
- The text needs to be without html, so I first rip the html with Html to text action
- Then I created scope for the prompts, and because these might fail I will add the following action configuration to run even if this scope is failed
- Calling the custom prompt with email body
- Adding the ChatGPT response as comment for the created backlog task
- Repeating this for the next custom prompt
Using the custom prompt is very straightforward, use the Create text with GPT using a prompt action, pass the plain text and wait for response. You could give here more instructions for the prompt like what is the context or what kind of output you wait for.
Then create HTTP Request to Azure DevOps for creating the comment. Repeat the same steps for the Seek for solution prompt.
Finally you have GPT answers as comments in your task. I think there should be better formatting, since this does not understand new line feeds. You can finetune this the way you find the most suitable in your case.
Solution blog post collection
- How to Streamline Customer Requests and Improve Consultant Productivity
- Use Power Automate to automate the creation of tasks from customer emails requests
- Add attachments and pictures to Azure DevOps tasks from email
- Use AI Builder to analyse the request problem and create solution for it
- Use Teams group chat for notifying developers of new tasks in backlog